Nonlinear Dyn (2016) 84:1637–1653

DOI 10.1007/s11071-015-2593-0

ORIGINAL PAPER

Application of dynamic optimisation to the trajectory of a

cable-suspended load
Łukasz Dra˛g

Received: 19 May 2015 / Accepted: 28 December 2015 / Published online: 14 January 2016
© The Author(s) 2016. This article is published with open access at Springerlink.com

Abstract The paper discusses a dynamic model of the or raisers, which frequently occur in the offshore engi-
system consisting of an on-board hoisting winch on a neering systems, have to account for their large sags
moving vessel, cable, and load. The model account- or bowing, variable configuration, and loads caused by
ing for large rope sag and hydrodynamic drag force water impact (wavy motion, buoyancy force, the pres-
was used to solve the problem of dynamic optimisa- ence of sea currents, hydrodynamic drag, volume of
tion. The essence of which is what angle of rotation associated water, etc.). To divide the flexible structure
should be selected for the hoisting winch to ensure that into discrete elements, the following methods are used:
the load during the defined movement of the vessel finite element method [1,2], lumped mass method [3],
shall remain at the set distance from the seabed, which segment method [4–6]. Generally speaking, for mod-
may be undulating. To solve this problem, the nonlinear elling of flexible structure in the offshore engineering
optimisation methods were used. The presented calcu- systems the methods used include those developed,
lation results show the efficiency of the developed 3D inter alia, in the analysis of multibody system dynamics
system model and its possible application to solve other with recognition of flexibility. A broad review of meth-
similar optimisation problems. ods used is presented in the paper [7]. For modelling
of flexible structures the rigid finite element method
Keywords Application · Dynamic model · Nonlinear (RFEM) is also used [8–10]. The modification of that
optimisation · Load positioning method is presented in this paper. A similar approach
was applied in the study [11] where the flexible seg-
ment model was presented. The model developed there
1 Introduction assumes, however, dependency of coordinates of ith
element on the coordinates of preceding elements. This
In contemporary offshore engineering systems, increas- assumption is redundant in the proposed RFEM modi-
ingly complicated devices are used that enable opera- fication due to the formulation of appropriate equations
tion at great depths and in difficult conditions. To simu- of geometric bonds. The issue of method selection used
late operation of such devices, complex dynamics mod- for dividing susceptible members such as cables or rais-
els, applied in commercial software packages (Rifflex, ers, into discrete elements, gains special importance
Orcaflex, Proteus), are required. The models of cables in solving the problem of controlling the propulsion
drives of marine equipment. The numerically effective
Ł. Dra˛g (B)
University of Bielsko–Biała, ul. Willowa 2,
model of the system dynamics, reflecting its character-
43–309 Bielsko–Biała, Poland istics with acceptable accuracy, enables the formulation
e-mail: [email protected] and even solution of appropriate optimisation tasks or

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1638 Ł. Dra˛g

their sequence for a specific set of control parameters. models presented in [16,17]. Similarly to the approach
The results may then be provided as inputs for artificial from [17], the longitudinal flexibility is omitted, which
neural networks and consequently used for control. The enables the equations of motion to be integrated using
questions of dynamic optimisation of cables and raisers relatively large integration step. So, the presented mod-
are quite rarely presented in the literature. This problem ification of RFEM is characterised by high numerical
for cranes operated in marine conditions is discussed in efficiency, which enables its use for dividing the cable
papers [12,13]. In the above papers, as in many others, into discrete elements while solving a dynamic opti-
the cable is modelled as an elastic and damping element misation problem. The dynamic optimisation problem
without mass. Such a simplification is not used in the considered further in this paper is shown in Fig. 1.
aforementioned commercial software packages. That It involves determination of the angle of rotation of
method was also applied to partition the raiser into dis- a hoisting winch ϕ(t), which shall ensure that the load
crete elements [14]. Also the monograph [15] formu- is at a constant distance from the surface of the sea
lated the problem of dynamic optimisation involving bottom.
the compensation of wave motion during pipe laying
using a drum method with the application of rigid finite
element method (RFEM). 2 Susceptible element model
This paper presents a 3D model of the elastic ele-
ment with bending and possibly torsional compliance, Figure 2a shows elastic beam element of constant cross
enabling the mapping of large cable and raiser sags. section, which during primary division was partitioned
The model developed through the application of own into n equal sections (Fig. 2b) with the length of:
modification of RFEM eliminates translation rigidity
from further consideration (shearing in both directions L
= (1)
and longitudinally one) and optionally torsion. The for- n
mulation of the rigid finite element method presented During secondary division, in the middle of these sec-
in this paper differs considerably from the previous for- tions spring damping elements were placed (sde) ⊗ as
mulations. The motion of each rigid finite element (rfe) shown in Fig. 2c.
is described by three displacements, two rotation angles Parts of unit located between unit ends and sde, and
reflecting bending, and one rotation angle reflecting tor- between sde, are hereinafter regarded as rigid elements
sion (when necessary). Thus, each rfe has either five or (rfe). Their number is n + 1 (numbered from 0 to n).
six degrees of freedom. The elements are connected by As a result of that division, the unit is replaced with
means of three geometrical constraint equations due to the n + 1 system of rigid solids (rfe) assuming mass
the necessary continuity of displacements at the points properties of the unit and n non-dimensional spring
connecting the elements. Such an approach has already damping elements (sde) without mass assuming unit’s
been applied by the author but for planar systems with elastic (bending) properties.
consideration of bending [16,17] or bending and lon- The operational procedure when the unit is of vari-
gitudinal flexibilities. In the formulation presented in able section or concentrated loads are attached is
[16], each rfe has four degrees of freedom: two transla- described in detail in papers [9,18].
tions, bending rotation, and elongation, while the ele- The movement of each rfe is defined by the vector
ments with three degrees of freedom (two translations components:
and bending) are used in [17]. Constraint equations are ⎡ ⎤ ⎡ ⎤
xi qi,1
formulated for both the planar and spatial models; how- ⎢ yi ⎥ ⎢ qi,2 ⎥
ever, there are 2*(n+1) constraint equations in planar ⎢ ⎥ ⎢ ⎥
⎢ z i ⎥ ⎢ qi,3 ⎥
models (2D) described in [16,17], while for the spatial ⎢
qi = ⎢ ⎥ = ⎢ ⎥ ⎢ ⎥, (2)
⎥
models (3D) considered in this paper, 3*(n+1) con- ⎢ ψi ⎥ ⎢ qi,4 ⎥
⎣ θi ⎦ ⎣ qi,5 ⎦
straint equations have yet to be formulated. In both for-
ϕi qi,6
mulations, n+1 is the number of rigid finite elements
into which the flexible link is divided. where xi , yi , z i are coordinates of point Ai and ψi , θi ,
Thus, the models of spatial slender systems pre- ϕi are Euler’s angles Z  Y  X  , presented in Fig. 3, that
sented in this paper are generalisations of the planar are rotations around system axis {i} combined with rfe

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1639

Fig. 1 Selection of angle of

rotation of a hoisting winch

Fig. 2 Division of an
elastic unit to spring
damping elements (sde) and
rigid finite elements (rfe), a
unit length, b primary
division, c secondary
division

Fig. 3 Rigid finite element

and its global coordinates

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1640 Ł. Dra˛g

i. Such angles are used in aviation mechanics and are Following the procedure referred to in (17), the
called heading, attitude, and bank [19]. Lagrange operators from kinetic energy of rfe i may
While applying homogeneous transformations, the be presented as follows:
coordinates of the point defined in the local coordinate d∂ E i ∂ Ei
system {i} combined with rfe i may be expressed in εqi (E i ) = − = Ai q̈i + hi , (6)
dt∂ q̇i ∂qi
the inertial system according to the following formula:
where ai, jk = tr {Bi, j Hi Bi,k
T } j, k = 1, ..., m,
ri = Bi ri , (3)
m m
where h i, j = q̇i,k q̇i,l tr {Bi,k Hi Bi,kl
T
} j = 1, ..., m,
 T
r i = x  y  z  1 —coordinates of the point in the k=1 l=1

local system {i} , m = 6—number of generalised coordinates describing

 T
ri = x y z 1 —coordinates of the point in the movement of rfe i.
global system {}, The calculation of Ai matrix elements and hi vec-
Ri ri tor using formulae referred to in (6) is very time-
Bi = 0 1
—homogeneous transformation consuming in terms of computer processing time. Once
matrix. derivatives of Bi matrix have been calculated, it would
⎡ ⎤⎡ ⎤ be desirable to calculate products and traces of matrices
cψi −sψi 0 cθi 0 sθi
Ri = ⎣ sψi cψi 0 ⎦ ⎣ 0 1 0 ⎦ present in formulae of ai, jk and h i, j . The high number
0 0 1 −sθi 0 cθi of elements present in (5) is at the same time equal to
⎡ ⎤ nil. Therefore, in this paper a similar method was used
1 0 0 to that referred to in [9]: products and traces of matrices,
× ⎣ 0 cϕi −sϕi ⎦ , referred to in (5), were calculated analytically.
0 sϕi cϕi Individual rigid finite elements are combined to
⎡ ⎤
xi form a kinematic chain by means of geometric bonds
ri = ⎣ yi ⎦ , cψi = cos ψi , sψi = sin ψi , (Fig. 4a). Forces present in links are shown in Fig. 4b.
zi To account for reactions in the links of rigid finite
cθi = cos θi , sθi = sin θi , cϕi = cos ϕi , elements (rfe), it is necessary to introduce generalised
sϕi = sin ϕi . forces. Generalised forces imposed on rfe i, originating
from Fi and from Fi+1 , might be presented as follows:
Elements of Bi matrix are defined unequivocally by the
components of the vector qi . The kinetic energy of rfe 
(A ) ∂ri T
i can be defined with following dependence [9]: Qi i = Fi , (7a)
∂qi
1  
E i = tr {Ḃi Hi ḂiT }, (4) (A ) ∂ri+1 T
2 Qi i+1 = · (−Fi+1 ) . (7b)
∂qi
where Hi = m i r T
i r i dm i .
It should be noted that elements of Hi matrix are Since
⎡ ⎤
constant. Under the assumption that axes y  , z  of the xi
{i} system are main axes, whereas xi axis is symmet- ri = ⎣ yi ⎦ , (8a)
rical axis, the Hi matrix referred to in (4) shall assume zi
⎡ ⎤ ⎡ ⎤
the following form: li xi
⎡ ⎤ ⎢ ⎥ ⎣ ⎦
h i,x 0 0 m i ai ri+1 = Ri ⎣ 0 ⎦ + yi , (8b)
⎢ 0 0 ⎥ 0 zi
⎢ h i,y 0 ⎥
Hi = ⎢ ⎥, (5)
⎣ 0 0 h i,z 0 ⎦ then
⎡ ⎤
m i ai 0 0 mi 1 0 0 0 0 0
∂ri ⎢ ⎥
where h i,x = x  i2 dm i , h i,y = y  i2 dm i , h i,z = = ⎣ 0 1 0 0 0 0 ⎦ = D, (9a)
mi mi ∂qi
2 0 0 1 0 0 0
m i z i dm i , m i —mass of the ith sde element.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1641

Fig. 4 Combination of
rigid finite elements (rfe)
into kinematic chain, a rigid
finite elements (rfe) and
spring damping elements
(sde), b forces in links

⎡ ⎤
1 0 0 −li sψi cθi −li sψi cθi 0 where
∂ri+1 ⎢ ⎥ ⎡ ⎤
=⎣ 0 1 0 li cψi cθi li sψi sθi 0 ⎦ = Di . −ln big(ψ̇n2 cψn cθn
∂qi ⎥
0 0 1 0 −cθi 0 ⎢
⎢ −2ψ̇n θ̇n sψn sθn + θ̇n2 cψn cθn ⎥
⎢  ⎥
(9b) ⎢ ⎥
Gn+1 = ⎢ −ln ψ̇n2 sψn cθn ⎥ + r̈n+1 .
⎢  ⎥
Finally, the generalised forces caused by forces affect- ⎢ −2ψ̇ θ̇ cψ sθ + θ̇ 2 sψ cθ ⎥
⎣ n n n n n n n ⎦
ing rigid finite elements rfe i reactions in Ai and Ai+1
may be noted as: ln θ̇n2 sθn
It should be noted here that given that the hoisting
Q0(F) = Q(A 0) T
0 = −D0 F1 , (10a)
winch drum rotates, the cable can reel/unreel on the
(Ai+1 )
Qi(F) = Qi(Ai ) + Qi drum. As a consequence, the length ln of the rfe n
T T changes and the variable figure of n, being the num-
= D Fi − Di Fi+1 , for i = 1, . . . , n. (10b)
  ber of elements into which the cable was divided.
li−1
The potential energy of terrestrial gravity forces of
Equations of bonds ri = ri−1 +Ri−1 0 connected
0 rigid finite elements (rfe i) may be presented as follows:
with the rigid finite elements (rfe i) link with i −1, after
Vg,i = m i g · yCi , (13)
their double differentiation, shall assume the following
form: where g—acceleration of gravity, yCi —y coordinate
Dq̈i − Di−1 q̈i−1 = Gi for i = 1, 2, ..., n, (11) of mass centre Ci .
Using the denotations in Fig. 2, the following calcu-
where lation can be made:
⎡  2 ⎤ ⎡ ⎤
−li−1 ψ̇i−1 cψi−1 cθi−1 − 2ψ̇i−1 0
⎢ ⎥
⎢ ×θ̇i−1 sψi−1 sθi−1 + θ̇i−12 cψ
i−1 cθi−1 ⎥
⎢ 1 ⎥
⎢ ⎥ ⎢ ⎥
⎢  2 ⎥ ⎢ ⎥
Gi = ⎢ −li−1 ψ̇i−1 sψi−1 cθi−1 − 2ψ̇i−1 ⎥. ∂ Vg,i ⎢ 0 ⎥
⎢  ⎥ = m i g ⎢ a cψ cθ ⎥ . (14)
⎢ ×θ̇ cψ sθ ⎥ ∂qi ⎢ i i i ⎥
i−1 i−1 + θ̇i−1 sψi−1 cθi−1 ⎦ ⎢ ⎥
2
⎣ i−1
⎣ −ai sψi sθi ⎦
li−1 θ̇i−1
2 sθ
i−1 0
Given that movement of point An+1 , rn+1 = rn+1 (t) is
When flexural (bending) and torsional compliance are
known, the additional equation of bonds may be noted
considered for discrete element, the moments caused
in acceleration-related form as:
by elastic strain of spring damping elements (sde) ⊗
− Dn q̈n = Gn+1 , (12) must be introduced to element motion equations. Sde

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1642 Ł. Dra˛g

Fig. 5 Spring damping element i, a moments generated by bending, b angles of relative strain

i generates moments Mi,ψ  , M  , and M  shown in A. In rfe i − 1 motion equations

i,θ i,ϕ
Fig. 5a.  
  
Generally, it is assumed [9] that moments Mi,ψ , Qi−1,Mi Mi,ψ , Mi,θ , Mi,ϕ
   
Mi,θ i Mi,ϕ can be nonlinear functions of angles ψi , = 0 0 0 Mi,ψ  
Mi,θ 
Mi,ϕ (18a)
θi i ϕi :
 
B. In rfe i motion equations
Mi,ψ = Mi,ψ (ψi ) , (15a)  
  

Mi,θ 
= Mi,θ (θi ) , (15b) Qi,Mi Mi,ψ , Mi,θ , Mi,ϕ

Mi,ϕ 
= Mi,ϕ (ϕi ) . (15c)  T T
  
= − 0 0 0 Mi,ψ Mi,θ Mi,ϕ (18b)
In the case of linear physical relationships, it can be  , M  , and M  may be roughly assumed
Figures Mi,ψ i,θ i,ϕ
assumed that:  , M  , and M  , or can be calculated
to be equal to Mi,ψ i,θ i,ϕ

Mi,ψ = ki,ψ ψi , (16a) precisely using the following formula:

Mi,θ = ki,θ θi , (16b) M i = RiT Ri−1 M i (19)

Mi,ϕ = ki,ϕ ϕi , (16c)
The motion equations of rfe 0, …, n (so far without
where ki,ψ , ki,θ , ki,ϕ are flexural and torsional rigidity taking into account the impact of aquatic environment)
factors. can be presented as follows:
Angles ψi , θi , and ϕi can be calculated based ⎧ T ∂ Vg,0
⎪
⎪ A0 q̈0 + D0 F1 = −h0 − ∂q0 + Q0,M1 = f 0
on the following formula: ⎪
⎪
⎪
⎪ ..
⎪
⎪ .
i
= Ri−1
T
Ri . ⎪
⎪
R (17) ⎪
⎪ T T
⎪
⎪ A q̈ − D Fi + Di Fi+1
 , M  , and M  are known, then their ⎨ i i
If moments Mi,ψ ∂V
i,θ i,ϕ = −hi − ∂qg,ii + Qi,Mi + Qi,Mi+1 = f i (20)
effect causes the occurrence of appropriate generalised ⎪
⎪
⎪
⎪ .
..
forces in the motion equations. ⎪
⎪
⎪
⎪
According to Fig. 5a, rigid finite elements (rfe ⎪
⎪ T T
⎪
⎪ An q̈n − D Fn + Dn Fn+1
i − 1) shall be affected by the moments Mi,ψ  , M ⎪
⎪
i,θ ⎩ = −h − ∂ Vg,n + Q
  , n ∂q0 n,Mn = f n
and Mi,ϕ , and rigid finite elements (rfe i) by -Mi,ψ
 , -M  . So spring damping element (sde) ⊗
-Mi,θ This is a system of 5·(n+1) differentiation ordinary
i,ϕ
of i number shall lead to occurrence of generalised equations of second order with 5·(n+1) unknowns,
forces: which are components of the vectors q0 ÷ qn , and
3·(n+1) unknowns, which are components of reaction

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1643

vectors F1 ÷Fn+1 . The above equations have to be sup-

plemented with 3·(n+1) bond equations (11) and (12):
⎧
⎪
⎪ Dq̈ − D0 q̈0 = G0
⎪. 1
⎪
⎪
⎪ ..
⎪
⎪
⎪
⎨ Dq̈ − D q̈
i i−1 i−1 = Gi
.. . (21)
⎪
⎪ .
⎪
⎪
⎪
⎪
⎪
⎪ Dq̈n − Dn−1 q̈n−1 = Gn
⎪
⎩
−Dn q̈n = Gn+1
(22) Fig. 6 Buoyancy force and hydrodynamic resistance
Before the presentation of the method how to solve
the system of equations (20) and (21) the water impact 3.2 Viscous resistance forces
forces were introduced into the system.
The circular unit section which is dξ in length is
3 Consideration of aquatic environment impact affected by the hydrodynamic resistance force [21]
(Fig. 6):
The aquatic environment impact forces, for examined df  H,i = S H,i (qi , q̇i, ξ )dξ, (25)
flexible elements, of large length and small cross sec-
where Dx ,D yz —longitudinal and lateral resistance fac-
tion, can be introduced using Morison equations [20].
tors, ⎡ ⎤
If forces caused by flow inside the element are dis- Vx
regarded, then consideration of aquatic environment V = ⎣ Vy ⎦ is relative velocity vector for dξ
impact requires taking into account the following:
Vz
– buoyancy force element in the coordinate system for rigid finite ele-
– viscous resistance forces ment rfe {i} ,
– force of inertia.  1

= V  y + V z .
2 2 2
Vyz
Below a method is presented for introducing the above
forces into the element motion equations. The generalised force that occurred due to the effect
of df H,i force corresponding to coordinate qi,k is pre-
sented by the following dependence:
3.1 Buoyancy force  li
Q H,i,k = dQ H,i,l
If ith rigid finite element (rfe) is fully immersed in 0
    
water, then the buoyancy force (Fig. 6) can be calcu- li ∂r T
= tr Ri S H,i (qi , q̇i , ξ ) dξ
lated as: ⎡ ⎤ ⎡ ⎤ 0 ∂qi,k
li 0 0 (26)
⎢ ⎥ ⎢ ⎥
fw,i = ⎣ gρw Ci ⎦dξ = ⎣ li ρw Ci ⎦ , (23)
The integrals present in (26) may be calculated using
0 0 0 Gauss formulae. The generalised forces generated by
where ρw is water specific gravity and Ci is cross- water resistance shall occur on the right-hand side of
sectional area for rigid finite element (rfe i). rigid finite elements (rfe) motion equations:
Meanwhile, the generalised force will be added in  T
Q H,i = Q H,i,1 , Q H,i,2 , Q H,i,3 , Q H,i,4 , Q H,i,5 , Q H,i,6 .
the motion equations for ith element:
⎡ ⎤ (27)
0
⎢ 1 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0 ⎥ 3.3 Force of inertia
Qw,i = gρw Ci ⎢ a cψ cθ ⎥ . (24)
⎢ i i i ⎥
⎢ ⎥
⎣ −ai sψi sθi ⎦ When applying the Morison equation, it may be
0 assumed that the force of inertia influencing ith rfe ele-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1644 Ł. Dra˛g

⎧ T
ment, which is dξ long, is defined in the inertial system ⎪
⎪ M0 q̈0 + D0 F1 = f0
⎪
⎪ ..
as: ⎪
⎪
⎪
⎨.
df I,i = [C M Ci ρw aw − (C M − 1)Ci ρw a] dξ T T
Mi q̈i − D Fi + Di Fi+1 = fi , (33)
⎪
⎪
= df I,i
w
+ df I,i
a
(28) ⎪
⎪ ..
⎪
⎪.
⎪
⎩ T
where Mn q̈n − Dn Fn+1 = fn
= ew where Mi = Ai − A I,i ,
I aw dξ, e I = C M Ci ρw ,
w w
df I,i
a
df I,i = eaI adξ, eaI = −(C M − 1)Ci ρw , fi = f i + Qw,i + Q M,i + QwI,i + QaI,i .
The equations of bonds (21) remain unchanged.
C M —coefficient, Fourth-order Runge–Kutta method was applied for
aw —water acceleration, integration of motion equations (33) with the constant
a—acceleration of rigid finite element (rfe i) with time step of numerical integration. Since bond equa-
the length of dξ . tions were transformed into acceleration-related form,
The first component in (28) shows forces caused by to stabilise them, the Baumgart method was used [22].
water movement (acceleration), and the second one by
movement (acceleration) of the rigid finite element dξ
rfe i. 4 Model validation
After df I,i integration over the length of element li ,
it may be stated that generalised forces caused by water Before proceeding to solving the optimisation prob-
acceleration are as follows: lem, the cable model was validated from the perspec-
 tives of static and dynamic calculations. A comparison
T
QwI,i = QwI,i,1 , QwI,i,2 , QwI,i,3 , QwI,i,4 , QwI,i,5 , QwI,i,6 was made of the results obtained using the model with
(29) accurate solution for catenary line, disregarding the
a
rigidity of flexible cable, and the results generated by
The generalised forces caused by effect of forces df I,i the Abaqus commercial software package. The impact
can be calculated as: of torsion on the results of numerical simulations and
 li 
a ∂r computation time was also examined.
QaI,i,k = tr df I,i , (30)
0 ∂qi,k
and after the following transformations: 4.1 Static problem
! "
QaI,i,k = tr B̈i Pi,a Bi,k
T
. (31) For static calculations, own results were benchmarked
⎡1 ⎤ against results obtained analytically. Exact values of x,
3 1 2
3 li 0 0 2 li z coordinates and force T in the cables were calculated
⎢ 0 0 ⎥
where Pi,a = ⎢
⎣ 0
0 0 ⎥. using the following formula [3]:
0 0 0 ⎦
Hs H V − wL + ws
1 2
2 li 0 0 li x(s) = + arcsinh
EA w H
The generalised forces, referred to in (31), may be
V − wL
presented as follows: − arcsinh , (34a)
H
= A I,i q̈i + h I,i .
QaI,i,k (32) s  ws 
! " z(s) = V − wL +
EA 2
where A I,i, jk = tr Bi, j Pi,a Bi,k
T , ⎧
⎪   2  2
1

H ⎨ V − wL + ws
m m ! " + 1+
h I,i, j = T
q̇i,k q̇i,k tr Bi,k Pi,a Bi,kl . w⎪ ⎩ H
k=1 i=1 1⎫
  2  2 ⎪
Taking into consideration the generalised forces orig- V − wL ⎬
− 1+ , (34b)
inating in the aquatic environment, one can record H ⎪
⎭
movement rfe 0 ÷ n in the following manner:

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1645

Fig. 7 Catenary line a assumed designations, b analytical solution and suggested method

 1
2 division of the rope chain into a higher number of ele-
T (s) = H 2 + (V − wL + ws)2 , (34c)
ments reduces computational error (Fig. 8b).
where w = ρ Ag—weight of cable unit length.
The numerical computations were made for the
cable with the length of L = 300 m, fixed cross section 4.2 Dynamics problem
A = 0.072 /4 m2 , Young modulus E = 1011 N/m2 ,
density ρ = 6.5 · 103 kg/m3, and loading with forces: In order to validate the dynamics model for large
H = 20000 N, V = 50000 N (Fig. 7a). The com- sags, the computational results were compared with
parison of the curves z(x) obtained for the analytical the results obtained using Abaqus commercial software
solution and for the proposed model, for the line divi- package. The calculations made using the commercial
sion n = 15 elements, is presented in Fig. 7b. As can be software package considered rope chain division into
seen, the proposed method correctly reflects the shape n=100 elements with the same characteristics as under
of the rope chain. static problem section. Then, load was applied to the
The values of relative percentage error εγ were cal- system as follows: end E of rope was subject to effect
 T
culated using the following formula: of forces F = Fx , Fy , Fz (Fig. 9). The curves of Fx
& &
& f c (si ) − f a & and Fy forces are presented in Fig. 9b, and the compo-
& γ γ &
εγ = & & × 100 % (35) nent Fz = −3000 N was assumed to be constant.
& f γa &
Numerical simulations were started in the position
where f γc —approximate value using the presented of free sag of rope. Total analytical time was 100 s.
method, f γa —value calculated using formulae (34a) Figure 10 shows the trajectory of E point in the x y
' plane.
and (34b), for si = ij=0 l j and i = 0...n, γ ∈ {x, z}.
Similarly to the static problem, very high confor-
Maximum relative percentage error εγmax was calcu-
mity was attained between own model results and those
lated using the following formula:
obtained from commercial package. Maximum rela-
εγmax = max εγ (si ) (36) tive error for displacement towards x and y for the
1≤i≤n
rope divided into n = 100 elements is less than 0.5 %.
The analysis of the values of errors shown in Fig. 8a The total simulation time using fourth-order Runge–
showed that the maximum differences in displacements Kutta method after the adoption of integration step
in x, z directions did not exceed 0.5 %. The highest h = 2 · 10 − 4 s was 5242 s. The fact that obtained
values of errors εx , εz are present at points where rope results are highly compliant with those of Abaqus com-
chain was characterised by the biggest curvature. The mercial software package proves that the rope model

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1646 Ł. Dra˛g

Fig. 8 Relative percentage errors a εx , εz , b εxmax ,εzmax

Fig. 9 Analysed system a

forces applied at the end of
rope, b curves of Fx and Fy
forces

proposed in this paper may be used in the dynamic G Ii0

ci,ϕ = , (37)
optimisation. 
The impact of torsion on the E point trajectory and where G—modulus of volume rigidity; Ii0 —polar
calculation time was examined. An important advan- moment of inertia for cross-sectional area; —length
tage of the model presented earlier is that omission of of initial element as shown in Fig. 1. Given small val-
torsion requires only the omission of ϕi (q1,6 ) coordi- ues of mass moments of inertia for rigid elements (rfe)
nate in the formula (2) and then reduction of matrix of the rope, high values of ci,ϕ lead to the occurrence
and vectors used in Eq. (21) and (33) from 6 × 6 and of high frequency of torsional vibrations. Meanwhile,
6 × 1 to 5 × 5 and 5 × 1. It should be also noted that this necessitates the reduction of the step of integration.
omission of shearing would enable drastic reduction Thus, the elimination of coordinates ϕi enables increase
of the integration step. For flexible elements such as a in step of integration of system displacement equation,
rope, the torsional rigidity factor is calculated using the and thus reduction of calculation time. Table 1 shows
following formula: the impact of the length of step of integration on cal-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1647

Fig. 10 E point trajectory

with a projection to x y
plane

Table 1 Displacement of the rope end E according to the proposed modification of RFEM including torsion and excluding torsion
t (s) Including torsion Excluding torsion

h = 1 × 10−4 h = 1 × 10−4 h = 1 × 10−3 h = 1 × 10−2 h = 2 × 10−2

x E (m) y E (m) x E (m) y E (m) x E (m) y E (m) x E (m) y E (m) x E (m) y E (m)

10 0 44.048 0 44.048 0 44.048 0 44.048 0 44.048

25 96.096 70.383 96.095 70.386 96.095 70.385 96.093 70.384 96.092 70.385
50 117.082 60.097 117.072 60.085 117.072 60.085 117.071 60.084 117.072 60.084
75 95.417 26.443 95.314 26.418 95.314 26.418 95.317 26.417 95.317 26.417
100 117.914 47.198 117.907 47.140 117.907 47.140 117.910 47.140 117.908 47.140

culation results (coordinates x E , y E ) under torsion and 5 Optimisation problem: stabilisation of rope end
torsion omission scenarios. at set depth
It may be seen that the E point trajectory error even
with the step of integration 200 times greater does not The system model consists of the drive system drum
exceed 0.15 %. Therefore, the torsion was omitted in and the rope. It was assumed in the optimisation that
the optimisation problem analysis shown below. vessel displacement at the place where rope hoist-

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1648 Ł. Dra˛g

where x0 , y0 , k, p, q are parameters of the function

describing sea bottom profile.
Calculation of the value of functional involves
integration of motion equations (33) for adopted func-
tions defining x A (t), y A (t). To analyse the problem as
nonlinear optimisation, it was assumed that ϕ(t) func-
tion is defined as third-degree catenation function. The
catenation function ϕi = ϕ(ti ) values are searched in
equidistant points of the 0; T  interval:
T
ti = i × for i = 1, . . . , p, (41)
p
where T —motion simulation time, p—number of
subintervals in 0; T .
It is assumed that value ϕ0 = ϕ(0) = 0 is known
and that it results from the system initial configuration.
The objective of minimisation of the function referred
Fig. 11 Model of the system used for dynamic optimisation to in (39) is to search for such values ϕ1 ÷ ϕ p being
 T
components of vector f= ϕ1 , ϕ2 , ..., ϕ p which fulfil
ing winch is located would be known. Assuming that limitations ϕi,min ≤ ϕi ≤ ϕi,max and minimise function
R radius of the drum is small, it would enable cal- = (f) referred to in (39).
culation of coordinates x An+1 = x A (t), y An+1 = To solve the problem of nonlinear optimisation, the
y A (t), z An+1 = z A (t). Because hoisting winch drum creeping simplex method was used. The motion equa-
may rotate, the rope’s length varies, which enables tions were integrated assuming the step of integration
adjustment of load position (height δ). In the exam- h ≤ 0.1 s securing proper stability and accuracy of
ined problem of dynamic optimisation the course of solution.
ϕ(x A (t), y A (t)) = ϕ(t) drum rotation was selected,
which shall ensure that despite horizontal movement
6 Results of numerical simulations
of the vessel, point A0 shall stay at the same distance
δ from the sea bottom (Fig. 11). The vertical displace-
In order to depict displacements towards x axis of E
ments of point A were omitted. As shown by the results
end of the rope sitting on motionless drum of the on-
published in [15], the vertical movement may be com-
board hoisting winch on the vessel moving towards x,
pensated by adoption of:
ϕ = ϕ(z A ) + ϕ(x A , y A ), (38)
Table 2 Rope and water parameters
zA
where ϕ(z A ) = rD , rD is radius of winch drum.
Element Parameter Value Unit
The objective function was defined as follows:
⎡ ⎤ 21 Line Length L 300 m
T
1  2 Young modulus E 1011 N/m2
=⎣ z A0 + z sb − z(x, y) − δ dt ⎦ , (39)
T Density ρ 6500 kg/m3
0 Diameter d 0.07 m
where z(x, y)—function describing seabed topograph- Number of division n 15 –
ical features,
 & z sb —seabed
 depth for A0 at t = 0, Water Density ρw 1025 kg/m3
z sb = − z A0 &t=0 − δ .
Dx coefficient 0.1 –
The function describing seabed topographical fea-
D yz coefficient 1.0 –
tures was defined as follows:
(     ) C M coefficient 2.0 –
(x −x0 ) 2 (y − y0 ) 2 [000]T m/s2
z(x, y) = k exp − − , Accelerationaw
p q Distance δ from the seabed 5 m
(40)

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1649

Fig. 12 E point trajectory

Fig. 13 Vessel movement—point An+1 in x and y directions a velocity, b vessel trajectory (point An+1 )

numerical simulations for data shown in Table 2 were Total simulation time was T = 210 s. The values of
made. As a result of hydrodynamic resistance having decision variables in the dynamic optimisation prob-
impact on the rope, the end of rope rises (Fig. 12). lem were selected at equidistant points, dividing 0; T 
The selection of the angle of rotation of the hoisting interval into p = 14 subintervals. It was assumed that
winch drum ϕ(t) in the problem of stabilisation of the motion of point An+1 (vessel) along x and y axes is
rope end at a set height above seabed shall concern the described as in Fig. 13a. Thus, the trajectory is the same
compensation of the rope end lifting caused by vessel as in Fig. 13b.
movement and elimination of changes caused by set Two options for seabed profile were examined: A—
seabed profile. flat (k = 0 in formula (40)), B—hill with the height of
It should be noted that when hoisting winch is immo- k = 15 m; for p = 20, q = 20, x0 = 70, y0 = 20.
bilised, the E line end is displaced along z axis by Consecutive drawings 14 ÷ 16 show E point trajec-
approximately 7 m. This is caused by the fact that the tory, before and after optimisation for both examined
rope length is maintained unchanged, despite the effect options of seabed profile under the assumption that the
of hydrodynamic resistance. rope end is kept at a set distance δ from the seabed.

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1650 Ł. Dra˛g

Fig. 14 E point trajectory before and after optimisation in the following planes: a xy, b xz, c yz

The value of objective function after optimisation, good, given the height at which E point of rope is raised
for A and B options, was A = 0.076 and B = 0.105, as a result of vessel movement (dotted lines in Fig. 14)
respectively, which means that the average deviation (Figs. 15, 16).
of the rope end from set distance δ = 5 m above the The range of angles of the rotation of the hoisting
seabed was below 0.11 m. The result of optimisation winch drum ϕ(t) selected during optimisation process
for examined seabed profiles should be considered as for option A and option B is shown in Fig. 17. In both

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Application of dynamic optimisation to the trajectory of a cable-suspended load 1651

Fig. 15 E point trajectory

before and after
optimisation—option A

Fig. 16 E point trajectory

before and after
optimisation—option B

cases, the angle of drum rotation was selected in such angle ϕ(t)), the distance is maintained at almost con-
manner as to eliminate changes caused by the raising stant level—approximately δ = 5 m above seabed.
of E end of the rope due to hydrodynamic resistance The results of dynamic optimisation presented above
and change of seabed profile. indicate that through appropriate selection of the hoist-
Shape and position of the rope end at selected points ing winch drum rotation angle, it is possible to control
in time depending on seabed profile are presented in rope end position, which will guarantee keeping it at
Fig. 18. To describe vessel movement in Fig. 13, it a set level above seabed. The major factor influencing
was noted that rope end rises maximally approx. 12 m the optimisation process time is the number of deci-
from seabed. Following the application of optimisa- sion variables. Therefore, it has been planned to subse-
tion process (selection of the hoisting winch rotation quently develop an algorithm ensuring the appropriate

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1652 Ł. Dra˛g

Fig. 17 Course of the hoisting winch drum rotation angle ϕ(t) before and after optimisation a option A, b option B

Fig. 18 Shape and position

of E point of the rope
before and after
optimisation for option B of
sea bottom profile at
selected points of time

123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Application of dynamic optimisation to the trajectory of a cable-suspended load 1653

selection of the quantity and distribution of points of 2. Jensen, G.A.: Offshore pipelaying dynamics. Doctoral thesis
division in the solution of the optimisation problem. at NTNU, Trondheim (2010)
3. Raman-Nair, W., Baddour, R.E.: Three-dimensional dynam-
ics of a flexible marine riser undergoing large elastic defor-
mations. Multibody Sys. Dyn. 10(4), 393–423 (2003)
7 Summary 4. Connelly, J.D., Huston, R.L.: The dynamics of flexible
multibody systems: a finite segment approach—I. Theoret-
The 3D model presented in the paper concerned with ical aspects. Comput. Struct. 50(2), 255–258 (1994)
5. Connelly, J.D., Huston, R.L.: The dynamics of flexible
the dynamics of flexible member with flexural compli-
multibody systems: a finite segment approach—II. Exam-
ance was developed through modification of the rigid ple problem. Comput. Struct. 50(2), 259–262 (1994)
finite element method. An important feature of the 6. Xu, X., Wang, S.: A flexible segment model based dynamics
model is a specific selection of generalised (global) calculation method for free hanging marine risers in re-entry.
China Ocean Eng. 26(1), 139–152 (2012)
coordinates. Each rigid element (rfe) has 5 or 6 DoF.
7. Dwivedy, S.K., Eberhard, P.: Dynamic analysis of flexi-
Unlike the conventional rigid finite element method ble manipulators, a literature review. Mech. Mach. Theory
(RFEM) where rigid elements are combined by means 41(7), 749–777 (2006)
of spring damping elements with 6 rigidity coefficients 8. Adamiec-Wójcik, I., Wojciech, S.: Application of a rigid
finite element method in dynamic analysis of plane manip-
(3 translational, 2 flexible, and 1 torsional), in this
ulators. Mech. Mach. Theory 28(3), 327–334 (1993)
paper elastic and damping elements have only 2 or 3 9. Wittbrodt, E., Adamiec-Wójcik, I., Wojciech, S.: Dynamics
rigidity factors (2 flexural and sometimes 1 torsional). of Flexible Multibody Systems. The Rigid Finite Element
The identical displacements at rigid finite element links Method. Springer, Berlin (2006)
10. Adamiec-Wójcik, I., Wittbrodt, E., Wojciech, S.: Rigid finite
were ensured by defining bond equations. This enabled
element in modelling of bending and longitudinal vibrations
elimination of translational rigidity coefficients of high of ropes. Int. J. Appl. Mech. Eng. 17(3), 665–667 (2012)
values from the system of motion equations. Conse- 11. Xu, X., Yao, B., Ren, P.: Dynamic calculation for underwater
quently, computation with a longer step of integration moving slender bodies based on a flexible segment model.
J. Ocean Eng. 57(1), 111–127 (2013)
was possible. The motion equations of the system were
12. Osiński, M., Wojciech, S.: Application of nonlinear opti-
supplemented with the parts connected with the impact misation methods to input shaping of the hoist drive of an
of marine environment. The model was validated by off-shore crane. Nonlinear Dyn. 17(4), 369–386 (1998)
comparison of the statics computations with analytical 13. Maczyński, A., Wojciech, S.: Stabilisation of load’s position
in offshore cranes. J. Offshore Mech. Arct. Eng. 134(2), 1–
solution (in catenary case), and dynamics computations
10 (2012)
were compared with the solution developed using MES 14. Wittbrodt, E., Szczotka, M., Maczyński, A., Wojciech, S.:
commercial software package. In both cases, satisfac- Rigid Finite Element Method in Analysis of Dynamics of
tory conformity of results was obtained already when Offshore Structures. Springer, Berlin (2013)
15. Szczotka, M.: Metoda sztywnych elementów skończonych
flexible member was divided into n ≥ 10 elements.
w modelowaniu nieliniowych układów w technice morskiej.
High numerical efficiency of the developed model and Wydawnictwo Politechniki Gdańskiej, Gdańsk (2011)
computer software enabled their application to solve 16. Adamiec-Wójcik, I., Brzozowska, L., Dra˛g, Ł.: Dynamics
the problem of dynamic optimisation. It is expected of flexible risers during vessel motion by means of the rigid
finite element method. J. Ocean Eng. 106, 102–114 (2015)
that the developed model and computer software shall
17. Dra˛g Ł.: Model of an artificial neural network for payload
be used in simulation of dynamic positioning of risers. positioning in sea waves. J. Ocean Eng. (In review, 2015)
18. Kruszewski, J., Gawroński, W., Wittbrodt, E., Najbar, F.,
Grabowski, S.: Metoda sztywnych elementów skończonych
Open Access This article is distributed under the terms of the (The Rigid Finite Element Method). Warszawa, Arkady
Creative Commons Attribution 4.0 International License (http:// (1975)
creativecommons.org/licenses/by/4.0/), which permits unrest- 19. Bauchan, O.A.: Flexible Multibody Dynamics. Solid
ricted use, distribution, and reproduction in any medium, pro- Mechanics and Its Applications, vol. 176. Springer, Berlin
vided you give appropriate credit to the original author(s) and (2011)
the source, provide a link to the Creative Commons license, and 20. Chakrabarti, S.: Handbook of Offshore Engineering. Else-
indicate if changes were made. vier, Amsterdam (2005)
21. Jensen, G.A., Safstromc, N., Nguyen, T.G., Fossen, T.I.:
A nonlinear PDE formulation for offshore vessel pipeline
References installation. J. Ocean Eng. 37, 365–377 (2010)
22. Baumgarte, J.: Stabilisation of constraints and integral of
1. Shabana, A.A.: Dynamics of multibody systems. Cambridge motion in dynamical system. Comput. Methods Appl. Mech.
University Press, Cambridge (1998) Eng. 1, 1–16 (1972)

123

Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”),
for small-scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are
maintained. By accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use
(“Terms”). For these purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or
a personal subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or
a personal subscription (to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the
Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data
internally within ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking,
analysis and reporting. We will not otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of
companies unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that
Users may not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to
circumvent access control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil
liability, or is otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by
Springer Nature in writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer
Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates
revenue, royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain.
Springer Nature journal content cannot be used for inter-library loans and librarians may not upload Springer Nature journal
content on a large scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any
information or content on this website and may remove it or features or functionality at our sole discretion, at any time with or
without notice. Springer Nature may revoke this licence to you at any time and remove access to any copies of the Springer Nature
journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express
or implied with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or
warranties imposed by law, including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be
licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other
manner not expressly permitted by these Terms, please contact Springer Nature at

[email protected]